One hundred thirty-three consecutive episodes of Pseudomonas aeruginosa bacteremia were prospectively followed in a university hospital over a 36-month period. The attack rate was 1.8 episodes per 1, 000 discharges, and 85% of the episodes were hospital acquired. P. aeruginosa bacteremia represented 13.6% and 25.6% of the episodes of nosocomial bacteremia and gram-negative nosocomial bacteremia, respectively. The crude mortality rate was 50%. A stepwise logistic regression analysis defined four variables as independently influencing the outcome: Development of septic shock (P =.00002), a granulocyte count <500/mm3 (P =.0008), inappropritate antibiotic therapy (P =.001), and the development of septic metastasis (P =.003). Among them, only the antibiotic treatment is easily amenable to medical intervention in order to improve the prognosis. Consequently, major efforts should focus on prevention and on the development of other therapeutic measures apart from antibiotic treatment.